LNG – Metrology for LNG - FIELD TEST OF A …...1. Laser generated at Base Unit 2. Laser guided to...

Shell Projects and Technology

FIELD TEST OF A RAMAN APPLICATION FOR LNG

Hans van der Poort, Shell Projects and Technology

Scott W. Brians, Scientific Instruments, Inc.

1 oktober 2011


RAMAN PHYSICS

oktober 2011 2

Unique Bonding Structure Vibra tiona l Mode


RAMAN SCATTERING

oktober 2011 3

∇ν = - 1

λi λs

1 Vibra tion


RAMAN SPECTROSCOPY

oktober 2011 4

Laser Analyte

Detector

∇ν = - 1

λi

λs

λs

1

λi


RAMAN SPECTROSCOPY EQUIPMENT

oktober 2011 5

Base Uni t Ana l y zer

Pr obe

In-S

itu

LNG

Mea

suri

ng

Poin

t

Cab l e

Fi ber Op t i c

Measurement Steps 1. Laser generated at Base Unit 2. Laser guided to Probe 3. Laser enters LNG at Probe Tip 4. Raman Scattering occurs in LNG 5. Scattered Light collected by Probe 6. Scattered Light guided to Detector in Base Unit 7. Analysis of Specie and Relative Abundance


RAMAN BASICS

oktober 2011 6

Laser frequency is important – sensitivity

proportional to ν4


RAMAN BASICS

oktober 2011 7

Stokes Rayleigh Anti-Stokes

E0 v=3 v=2 v=1 v=0

Virtual state

hvex hvex hvex hvex

h(vex-vv) h(vex+vv)

E1


RAMAN COMPARED TO MID-IR

oktober 2011 8

Raman spectroscopy provides information on the chemical make-up of molecules by observing the vibrational energies of the molecules.

Raman is complementary to mid-IR BUT different intensities and selectivity.

Raman spectra tend to be less cluttered than IR, much less affected by water.

The Chemical Specificity of mid-IR, but with the ease of implementation and sampling of NIR!


LNG RAMAN SPECTRUM

oktober 2011 9

500 1000 1500 2000 2500 3000

0

2

4

6

x 105

Raman Shift (1/cm)

Rel

ativ

e In

tens

ity

2800 2900 3000 3100

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

x 106

Raman Shift (1/cm)

Rel

ativ

e In

tens

ity

500 1000 15000.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

x 104

Raman Shift (1/cm)

Rel

ativ

e In

tens

ity

2200 2300 2400 2500 2600 2700

1

1.5

2

2.5

3

x 104

Raman Shift (1/cm)

Rel

ativ

e In

tens

ity

Ethane 2.24-10.24%

Propane 0.39-3.39%

Nitrogen 0.10-0.46%

Methane 85.06-97.06%

i-butane 0.39-3.39%


WHY USE RAMAN SPECTROSCOPY?

oktober 2011 10

Combines advantages of NIR and mid-IR Sample through common windows / conta iners

glass, sapphire, polymer Utilize fiber-optics Direct insertion of immersion probes into process Fast ana lysis No sample prepara tion Aqueous Samples – easy Specificity for robust methods Univaria te or multivaria te ca libra tions The only? technique tha t can quantify dia tomics

without removing sample from process

RXN2

RXN4


TEST LOCATION

Test location at Hazira LNG Terminal in India.

History

Started with manua l sampling (2005)

Piston sampler with GC (2008)

Raman testing

Reason to select Hazira

Wide range of ca rgoes

oktober 2011 11


INSTALLATION

The Raman spectroscope, holding the laser source is installed in the Jetty substation located over 300 meters from the sampling point.

Sample probe is installed next to the off-take for the LNG sampling system in order to be able to properly compare the results of the Raman against the online GC.

oktober 2011 12


INSTALLATION

oktober 2011 13


QUALITY ASSURANCE

The repeatability of the online GC was determined during the initial phase of the Raman test.

For this a certified calibration gas was analyzed 60 times with the online GC.

The repeatability was calculated using the algorithm as described in ISO 6974-5.

All repeatability's for the individual components did meet the international criteria of both ISO 6974 and GPA 2261 for all components.

oktober 2011 14


QUALITY ASSURANCE

Repeatability result on Methane for online GC

“R” is based on ISO 6974-5

oktober 2011 15

89.28

89.3

89.32

89.34

89.36

89.38

89.4

89.42

89.44

89.46

89.48

89.5

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59

Measured

Cal gas

Upper limit R

Lower limit R


QUALITY ASSURANCE

Online versus offline GC

Additiona lly the online and offline GC ana lyses results were compared

As criteria we have used the Reproducibility of ISO 6974-3 (offline) and ISO 6974-5 (online) (√(0,072+0,182)).

The delta plot shows tha t we do not have any significant difference between the labora tory and online samples.

oktober 2011 16


QUALITY ASSURANCE

Online compared to offline for Methane

oktober 2011 17


INITIAL RESULTS OF THE RAMAN ANALYSES

The initial results of the Raman test showed a significant offset in the composition compared to the online GC analyses data.

oktober 2011 18

88

88.5

89

89.5

90

90.5

91

91.5

00:00:00 01:12:00 02:24:00 03:36:00 04:48:00

Met

hane

in m

ol%

Time of analyses

Methane measurement GC vs Raman

Raman

GC



Initial reasons for problems

The vendor expected a number of reasons for the initia l significant devia tion of results.

Calibra tion of the fiber optics and probe.

Fouling on the optica l windows inside the spectrograph.

oktober 2011 19



Calibration of the fiber optics and probe

During the initia l insta lla tion of the Raman system, the vendor has to correct for any intensity losses caused by the fiber optics and the sample probe this was not properly done, an incorrect correction spectrum was used to subtract the noise of the fibers from the main Raman spectra .

Unfortuna tely this was done before the raw da ta was saved, making it impossible to reprocess the da ta .

oktober 2011 20



Fouling on the optical windows inside the spectrograph

An epoxy resin used to fix an optica l window tha t was insta lled just before the main CCD camera detector appeared to have released chemica l vapors. The vapors deposited on the CCD camera . This caused the Raman peaks in the spectrum to display a slight ta iling.

oktober 2011 21


NEW TEST RESULTS OF RAMAN ON GHV

oktober 2011 22

1020

1040

1060

1080

1100

1120

1140

4/Ju

n 4/

Jun

4/Ju

n 5/

Jun

10/J

un

10/J

un

10/J

un

11/J

un

11/J

un

14/J

un

14/J

un

14/J

un

14/J

un

14/J

un

22/J

un

22/J

un

23/J

un

23/J

un

23/J

un

2/Ju

l 2/

Jul

2/Ju

l 3/

Jul

3/Ju

l 19

/Jul

19

/Jul

20

/Jul

20

/Jul

20

/Jul

20

/Jul

20

/Jul

20

/Jul

23

/Jul

23

/Jul

24

/Jul

24

/Jul

4/

Aug

4/

Aug

5/

Aug

5/

Aug

5/

Aug

5/

Aug

5/

Aug

5/

Aug

Calo

ric

Val

ue [B

TU]

Energy Content of LNG by Online GC, Online Raman and Offline GC

GC Raman Offline


NEW TEST RESULTS OF RAMAN ON GHV

oktober 2011 23

The new test results have demonstrated the repeatability of the Raman on the GHV to be 0.020%* relative.

Relative standard deviation for the GHV is 0.20% using the conventional method (Raman 10 times better compared to the conventional custody transfer method).

No significant deviation in bias compared to the online GC (based on t-test).

*results based on one unloading (14-6) with 166 measurements, 2 outliers removed due to GC communication failure


MAIN CHALLENGES STILL TO BE TAKEN

oktober 2011 24

Although the tests are looking promising, the lower concentrations of the heavy hydrocarbons and nitrogen are not being detected properly .

This is due to the low concentrations (0-100ppm).



oktober 2011 25

Results of C5

0

0.005

0.01

0.015

0.02

0.025

Rela

tive

Abu

ndan

ce [%

]

n-Pentane GC

Offline

Raman

Full Scale = 1.0 BTU



oktober 2011 26

Results of Nitrogen

0

0.1

0.2

0.3

0.4

0.5

Rela

tive

Abu

ndan

ce [%

]

Nitrogen

GC Offline Raman



oktober 2011 27

With regards to the deviation in Nitrogen you can clearly see it is becoming more significant.

However it was expected that the GC-sampling valve was having some problems with internal leaking.

This caused a larger number of spikes in the initial data and especially showed on offset in the nitrogen.



oktober 2011 28

Possible solutions

In order to be able to more accura tely detect the lower levels of heavy hydrocarbons and nitrogen, it was proposed to average the Raman da ta over a longer time period for these components.

By doing this you will reduce the signa l to noise ra tio, improving the detection limit.



oktober 2011 29

Contribution of N2 on final heating value

In genera l the devia tion in the nitrogen concentra tions between the online GC and the Raman system results in an average devia tion in BTU of 0,99. In this the Raman measurement is a lways displaying a higher nitrogen content.

However it is questionable if this is caused only by the Raman.

Currently the average nitrogen devia tion is 0,03%.



oktober 2011 30

Nitrogen performance of online GC with calibration gas

0.48

0.485

0.49

0.495

0.5

0.505

0.51

0.515

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59

Measured

Cal gas


CONCLUSIONS

oktober 2011 31

Based on the current data, the Raman system has proven to be more stable, with a much lower repeatability compared to the conventional GC technique.

To a large extend this is because the LNG does not need to be vaporized first.

The Raman system needs specific knowledge during commissioning and start up.

Once operational little maintenance is required.


WHAT STILL NEEDS TO BE DONE

oktober 2011 32

The mean time between failure has not been determined yet. This will have to be determined at a real installation over a longer period of time (3 years minimal).

Temperature stability of the measurement has not been determined. For this the analyzer will have to be installed in a different environment that will experience larger temperature swings.


FUTURE NEEDS FOR RAMAN

oktober 2011 33

In order to get this technique accepted for online custody transfer, an international ISO/ ASTM/ GPA standard will have to be drafted.

Additionally the LNG industry will have to build large trust in the new technique.


Q & A

oktober 2011 34

LNG – Metrology for LNG - FIELD TEST OF A …...1. Laser generated at Base Unit 2. Laser guided to...

Documents

Transcript of LNG – Metrology for LNG - FIELD TEST OF A …...1. Laser generated at Base Unit 2. Laser guided to...